The information load (IL) analysis, first introduced forthe two-dimensional approach (Carlotti and Magnani,2009), is applied to the inversion ofMIPAS (Michelson Interferometer for Passive Atmospheric Sounding) observations operated with a 1-dimensional (1-D) retrieval algorithm. The ILdistribution of MIPAS spectra is shown to be often asymmetrical with respectto the tangent points of the observations and permits us to define thepreferential latitude where the profiles retrieved with a 1-D algorithmshould be geo-located. Therefore, defining the geo-location of the retrievedprofile by means of the tangent points leads to a "position error". We assess the amplitude of the position error for some ofthe MIPAS main products and we show that the IL analysis can also be used asa tool for the selection of spectral intervals that, when analyzed, minimize theposition error of the retrieved profile. When the temperature () profilesare used for the retrieval of volume mixing ratio (VMR) of atmosphericconstituents, the -position error (of the order of 1.5 degrees of latitude)induces a VMR error that is directly connected with the horizontal gradients. Temperature profiles can be externally-provided or determined ina previous step of the retrieval process. In the first case, the IL analysisshows that a meaningful fraction (often exceeding 50%) of the VMR errorderiving from the 1-D approximation is to be attributed to the mismatchbetween the position assigned to the external profile and the positionswhere is required by the analyzed observations. In the second case theretrieved values suffer by an error of 1.5–2 K due to neglecting thehorizontal variability of ; however the error induced on VMRs is of minorconcern because of the generally small mismatch between the IL distributionof the observations analyzed to retrieve and those analyzed toretrieve the VMR target. An estimate of the contribution of the -position error tothe error budget is provided for MIPAS main products. This study shows thatthe information load analysis can be successfully exploited in a 1-Dcontext that makes the assumption of horizontal homogeneity of the analyzedportion of atmosphere. The analysis that we propose can be extended to the1-D inversion of other limb-sounding experiments.
展开▼
